Fast obstacle detection using adaptive image segmentation and stereo vision

The algorithm firstly extracted average chroma of reference road in real time and segmented the images adaptively to detect the suspicious area primarily.Secondly,it detected the obstacle areas clearly using stereo vision on the diagonals of the suspicious obstacle areas.Proved the corresponding relationship of area's diagonals of right-and-left images when two cameras were parallel.This approach can adapt to the changes of illuminations automatically and improve the speed of the algorithm greatly.Experimental results show this method is valid and reliable.